A floater is usually plan B. It's a failure to correctly predict whether you can get to the rim and arguably a failure to correctly kick it out to another player to drive it again or shoot the open 3.
I think part of why floaters are generally a low percentage shot (it depends on the guy) is that most of these players have spent a lifetime being able to get to the rim and not having to resort to plan B too often. A floater is a useful shot and should be practiced by players with more marginal athletic ability (whether that be speed, strength, or height) who can't reliably get to the rim. Of course, the pull up jumper from 10 feet is also an option, and better still if the player learns to use the backboard when the physics (distance and angle)/shot blocking threats dictate.
But you still have to be able to make the shot at a pretty high percentage. Driving and getting fouled is great, layups are great, and in the absence of that you need a pretty good floater/jumper for that to be a better option than kicking out (time on the shot clock excepting, of course).
You can't check league wide stats on it, but NBA.com tracks shot types that you can check on individual player pages. I'm not sure what consistently distinguishes a driving floating jump shot from a floating jump shot (looking at the video of each attempt type proves to be of little assistance), but they track those. You'd have to go through the league by hand (or have a web crawling script compile the data) to get league wide stats.
Trae Young has a decent floater, for example. It's not a bad shot for him to take, especially relative to kicking it out to one of his teammates from last year.